When and how should biotic interactions be considered in models of species niches and distributions?

详细信息    查看全文

• 作者：Robert P. Anderson
• 刊名：Journal of Biogeography
• 出版年：2017
• 出版时间：January 2017
• 年：2017
• 卷：44
• 期：1
• 页码：8-17
• 全文大小：241K
• ISSN：1365-2699

文摘

Biotic interactions can influence the ranges and abundances of species, but no clear guidelines exist for integrating them into correlative models of niches and distributions. Niche/distributional models characterize environmental/habitat suitability or species presence using predictor variables unaffected by (= unlinked to) the population of the focal species. Such variables (termed ‘scenopoetic’) typically have been considered to include only abiotic factors. In contrast, population–demographic approaches model the abundance of the focal species by including linked predictor variables, which frequently are biotic interactors. Nevertheless, a focal species might hold no, or negligible, population-level effects on its biotic interactors. Hence, contrary to current theory, such interactors would represent unlinked variables valid and potentially very useful for niche/distributional models. Consideration of population-level effects indicates that facilitators and affecting amensals (species that negatively affect another species but are not affected by it) constitute unlinked variables, but commensals and affected amensals do not. For competitors, mutualists, predators/prey, consumers/resources, and parasites/hosts, additional information is necessary. Specifically, available ecological/natural history information for the particular species involved (e.g. regarding specificity) and theory regarding ecological networks can allow identification of interactors that are likely to be unlinked or nearly so. Including an unlinked biotic interactor as a predictor variable in a niche/distributional model should improve predictions when the effects of the biotic interactor vary across the study region, or in another place or time period. Other relevant interactors must be taken into account by post-processing a niche/distributional model, or via population–demographic models that require abundance data over time. This framework should improve current correlative models and highlights areas requiring progress.